Inspiration
Increasing problem of tracking power generation devices and their state over time. We want to be capable to give customer approximate forecast about expected energy generation and detect deviations.
What it does
Depending on the technical turbines or panels data and meteo, predicts power generation over time.
How we built it
Through performing a lot of data discovery, choosing efficient models for solving such a problem, by trying different approaches and views on the data and how it should be handled
Challenges we ran into
Lack of data, some data is not possible to find, inaccuracy of many data sources
Accomplishments that we're proud of
It works, it looks stylish, it has good results and we were not sleeping all night
What we learned
Better work organization, different data challenges and to deal better with such data distributions
What's next for EnergyBoost
Check more approaches and extend models via adding more data and trying models that we were not capable to check because of time limitations
Built With
- pandas
- python
- scikit-learn
- streamlit